Time-Identification in MemoryMuch of ordinary memory is autobiographical; memory of what one saw and did, where and when. It may derive from your own past experiences, or from what other people told you about your past life. It may be phenomenologically rich, redolent of that autumn afternoon so long ago, or a few austere reports of what happened. But all autobiographical memory is first-person memory, stateable using 'I'. It is a memory you would express by saying, 'I remember I . . .'.My question is whether this autobiographical memory requires the conception of time as linear. To think of time as linear is to suppose that times are ordered by 'earlier than' in a way that is irreflexive, anti-symmetric, transitive and connected. If we identify times using the ordinary Western clock and calendar we do seem to be supposing that time is linear, and we are using a method of identifying times that could be said to be 'canonical' in the sense that the temporal relations between any two times can simply be read off from the ways in which they are identified: it is immediately apparent that, for example, Thursday 25th December 1963 is before Friday 26th December 1963. But autobiographical memory could hardly be thought to depend upon the use of the clock and calendar. Presumably humans had autobiographical memory even before the invention of the clock and calendar, children now still develop autobiographical memory before they learn how to use the clock and calendar, and people in contemporary societies which do not use the clock and calendar at all may still have autobiographical memory. So if autobiographical memory involves the identification of past times, it must use a way of identifying times that is more primitive than the clock and calendar. My question then is whether in using this more primitive way of identifying past times we are taking it that time is linear. Other structures that might be assigned to the time of autobiographical memory are, for example, those involving a tree structure, or disjoint though individually linear time-streams.Psychologists who work on human memory are often hostile to the idea that autobiographical memory involves a conception of time as linear (Friedman 1990). One reason for this is that they want to resist the idea of an 'intrinsic temporal code', a kind of internal, innate analogue of the clock and calendar system, which would be canonical in the sense that the temporal relations between any two times identified using the code could simply be read off from the ways in which they were identified. There is no reason to suppose that there is such a